@audio/loudness-contrast

Speech contrast — foreground/background RMS difference, WCAG 2.0 SC 1.4.7 (Audacity Contrast)

npm install @audio/loudness-contrast

import speechContrast from '@audio/loudness-contrast'

Power-averaged foreground/background RMS in dB, per the Audacity Contrast analyzer — the reference tool for WCAG 2.0 Success Criterion 1.4.7 (Low or No Background Audio). Two modes:

• Explicit — pass fg/bg as [at, duration] second slices, mirroring Audacity's two-selection workflow.
• Auto — omit either slice and 10 ms frames are pooled into foreground/background by a dB threshold (an extension beyond Audacity, which always takes two manual selections). A slice given explicitly is excluded from the auto scan on the other side, so a selection never contaminates the opposite pool.

speechContrast(channels, { fg: [0, 1], bg: [4, 1] })   // explicit: two 1 s slices
speechContrast(channels)                                // auto: threshold-pooled

| Param | Default | | |---|---|---| | fs | 48000 | Sample rate, Hz | | fg | — | [at, duration] seconds — explicit foreground slice; omit for auto mode | | bg | — | [at, duration] seconds — explicit background slice; omit for auto mode | | threshold | -30 | dB level splitting auto-mode frames into foreground/background | | frame | 0.01 | Auto-mode classification frame, seconds (Audacity: 10 ms chunks) |

Accepts Float32Array (mono) or Float32Array[] (multichannel — RMS is power-averaged across channels before the dB conversion).

Returns { foreground, background, contrast, pass } — foreground/background in dB, contrast = foreground - background, pass = contrast >= 20 (the WCAG 2.0 SC 1.4.7 threshold).

Use when: validating that dialogue/narration is intelligible over background music or noise — captioning/accessibility pipelines, video QC.

Per WCAG 2.0 SC 1.4.7 / Audacity Contrast; ratio dB, not LUFS.

Part of @audio/loudness — the loudness family umbrella. Documented from the reference implementation.

MIT © audiojs